On Representational Music V. A. Howard Noûs, Vol. 6, No. 1. (Mar., 1972), pp. 41-53. Stable URL: http://links.jstor.org/sici?sici=0029-4624%28197203%296%3A1%3C41%3AORM%3E2.0.CO%3B2-K Noûs is currently published by Blackwell Publishing.
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/black.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.
The JSTOR Archive is a trusted digital repository providing for long-term preservation and access to leading academic journals and scholarly literature from around the world. The Archive is supported by libraries, scholarly societies, publishers, and foundations. It is an initiative of JSTOR, a not-for-profit organization with a mission to help the scholarly community take advantage of advances in technology. For more information regarding JSTOR, please contact
[email protected].
http://www.jstor.org Wed Aug 1 00:09:37 2007
SEVENTH SYMPOSIUM
On Representational Music
HARVARD PROJECT
ZERO,and
THE UNIVERSITY OF WESTERN ONTARIO
Commentators: ANNETTEBARNES
LEEBROWN
AMHERST COLLEGE OHIO STATE
UNIVERSITY
1. The Visual and the Musical1 Often the question whether music can be representational is broached as implying that music as an auditory event could become visual by synesthesis or substitute for a visual symbol. If, as Goodman argues, "Nothing is intrinsically a representation" ; but rather, "status as a representation is relative to a symbol system." ([2], p. 226), then it is misleading to ask whether music can usurp the function of pictures or even whether it is "by nature" nonrepresentational. T h e appropriate question is whether music can belong to a representational system as pictures more typically do. However, being a picture is neither necessary nor sufficient to being a representation as amply attested by non-representational styles of painting. "Objects and events, visual and nonvisual, can be represented by either visual or nonvisual symbols." ([2], p. 231) So it seems there is more to representation than meets the eye and may include what meets the ear in some musical contexts. It is the primary purpose of this paper to sketch the special conditions This paper was improved by the comments of Messrs. Nelson Goodman and T. Graham Roupas. I am grateful to the Canada Council for a grant in support of the research.
41
of auditory representation in so-called "pictorial" music while distinguishing musical representation from other programmatic symbol relations, particularly expression, description, and naming with which it is easily confused. One source of confusion is an ambiguous, common use of the word 'represents' roughly synonymous with 'symbolizes', 'stands for', or 'refers to' where the latter encompass many different symbol relations including pictorial representation. We shall be limited to consideration of the musical analogues of pictorial representation or depiction. T h e relative musical merits of attempts to use music to represent objects and events are not at issue. The technical notion of a representational system is Goodman's as described in Languages of Art and covers visual and auditory as well as other kinds of symbols. While this restricted sense of representation is suggested and in part controlled by ordinary uses of the word in aesthetic contexts, it does not correspond to all its aesthetic or derivative meanings anymore than such meanings correspond to each other. Pictorial representation, for instance, has little to do with parliamentary representation but more with musical or gestural representation. The terms 'expression', 'representation', and 'description' identify specific symbol relations found in both the visual and musical arts. I will argue that music can and often does meet the necessary and sufficient conditions of representation as well as of description and expression.
2. Onomatopoeia and Representation A natural place to begin an inquiry into musical representation is onomatopoeic sounds such as bird twitters, canon shots, water swirlings, drum rolls, wind howls, and the like. Examples such as these might tempt one to conclude that only those parts of music which are imitative replicas of "natural" sounds are representational. This would be a mistake; for, as I hope to show, onomatopoeia, though imitative, is often non-representational, while musical representation can be either imitative or not. There may be good reasons for avoiding or at least qualifying talk of imitation, but they can wait. I n the meantime, let me strike a comparison with Goodman's analysis of our ordinary use of the verb 'represents' as applied to pictures. (See [2], pp. 21-31) Goodman holds that to say something "represents" or is a "picture of" something else is ambiguous between saying what, if
ON REPRESENTATIONAL MUSIC
43
anything, the picture denotes and what kind of picture it is; that is, between (1) X is a picture of (i.e., denotes) Y; and (2) X is a Y-picture (or representation). I n sense (I), 'represents' or 'picture of' behaves as a dyadic predicate indicating what things are labelled by the picture. I n sense (2), 'represents' becomes a monadic predicate referring not to objects or events denoted by the picture but rather to how the picture itself is to be classified regardless of whether there is or ever was anything in the world corresponding to it. T h e difference is between the pictorial labelling of things and the labelling of pictorial things. Implicitly, we mark this distinction every time we notice a departure from the standard case where (1) and (2) hold simultaneously: where X is a picture of (denotes) Y, and X is a Y-picture (e.g., a Churchill-picture denoting Churchill). Picture X may well denote Y without being a Y-picture as in caricature or abstract symbolism (e.g., Christ represented by a fish), or, in the case of fictional beings, X may be a Y-picture without denoting anything. Onomatopoeia in musical works is the analogue of the standard case in pictorial representation. Roughly put,
(3) Passage S denotes R, and S is an R-sound (e.g., canon shots, church bells, etc.). This simply means that the work-sound exemplifies some of the same literal properties as some other sound which it denotes. Provisionally, this is all that is intended by saying that onomatopoeic work-sounds represent imitatively. All other instances of musical representation fall into the category of non-imitative worksounds where
(4) Passage S denotes R, and S is not an R-sound. Eventually, we shall have to distinguish other sorts of denoting but non-imitative passages-what I call musical names and descriptions-from representational ones. Just now, I should like to say a bit more about onomatopoeia and imitation.
3. Onomatopoeia and Imitation An important difference between onomatopoeia and the standard case of pictorial representation is in the kind and degree
of imitation. Onomatopoeic sounds in musical compositions are frequently thought to be replicas of other sounds, whereas a manpicture could hardly be a man-replica even if we are able to attach some significance to "looks exactly like a man". Further, a sound in onomatopoeia, if used denotatively, is a self-denoting label; that is, it denotes all sounds having the properties it exemplifies including itself. (See [2], pp. 61n and 81) I n this way, a bang-sound may become a surrogate for the predicate 'bang' by labelling other bang-sounds. By contrast, a man-picture of Trudeau denotes him but not itself. While I have spoken of onomatopoeia as imitative sound, imitation is no reliable criterion either of a sound being onomatopoeic or representational. Goodman observes that "The variability of exemplification is amusingly attested . . . by some linguistic curiosities: it seems that French dogs bark 'gnaf-gnaf' rather than 'bow-bow'; [one might add 'woof-woof' in English] that German cats purr 'schnurr-schnurr,' French cats 'ron-ron'; and in France a drip-drip is a plouf-plouf." ([2], p. 61n) Certainly there is as much variability among church bell sounds and bird twitter sounds as among onomatopoeic words in language. The fact that a sound is treated as onomatopoeic is at least as likely an explanation of its resemblance to other sounds which it denotes as its being a soundreplica of them. Furthermore, we do not usually think of successive toots of a train whistle or gull screeches as symbolizing, signifying, or denoting each other anymore than do exact replicas of coins or pieces of furniture. Similarity between two things in some or every respect does not imply that one denotes the other. (See [2], pp. 3-6; also [3], p. 19.) On the one hand, sounds dissimilar in many respects are treated as onomatopoeic, and on the other, there are sound-replicas which are not onomatopoeic. So even if onomatopoeic sounds were not as variable as they are, similarity or imitation would not explain the onomatopoeic labelling of one sound by another. Accordingly, neither can imitation explain any subset of musical denotation including representation. I t is noteworthy too that Goodman's examples of onomatopoeia above are all linguistic and therefore by definition non-representational; that is, they consist of verbal inscriptions belonging to systems that are syntactically disjoint and articulate. I hasten to add that I am not arguing that onomatopoeia is non-imitative, but rather that imitation is no explanation of onomatopoeic reference in representational or linguistic systems. Instead, it seems that the resemblances we discover between sounds can often be attributed to their being
ON REPRESENTATIONAL MUSIC
45
used onomatopoeically as much as to similarities antecedently perceived.
4. Scoring and Representing Towards the end of Languages of Art, Goodman caps his rejection of imitation as the basis of pictorial representation with these condensed remarks on musical denoting. One quite incidental consequence [of the analysis of pictorial representation] concerns representation in music. Here no more than in painting does representation require imitation. But if a performance of a work defined by a standard score denotes at all, it still does not represent; for as a performance of such a work it belongs to an articulate set. The same sound-event, taken as belonging to a dense set of auditory symbols, may represent. Thus electronic music without any notation or language properly so-called may be representational, while music under standard notation, if denotative at all, is descriptive. This is a minor curiosity, especially since denotation plays so small a role in music. ([2], p. 232)
I am inclined to believe that denotation plays a larger role in programme music and especially opera than Goodman's last sentence might suggest, but that is incidental. Two kinds of musical denoting are distinguished here: representation in a system that is syntactically dense, and description in one that is syntactically disjoint and articulate. For the moment, it is sufficient to mark the difference by whether the durations of pitches and silences of a performance are assignable to and identified by a score in standard notation. If pitches and durations are so notated, then score and performance are mutually recoverable: score from performance and conversely. (121, p. 178) I n neither representation nor description is the music uniquely recoverable from the objects or events denoted (e.g., a waterfall, battle, etc.), so that both are semantically non-disjoint, dense, or inarticulate. Thus, while musical representation shares certain symbol features in common with pictorial systems, musical description shares others in common with linguistic systems. Usually, pictorial systems are both syntactically and semantically dense, whereas linguistic systems are syntactically disjoint, articulate, and tend to be semantically dense. I t might appear from this that the mere existence of an appropriate score could mark the difference among denoting passages between those which represent and those which describe. However, that would be an arbitrary and nominal difference at best. I t is not just the presence or absence of a score that marks
the difference between the representational and the descriptive but rather scoreability. I t must be clear what are the characters and inscriptions of a notational scheme that make for duplication of the performance sounds. Performances which can be duplicated via such a scheme would also be said to belong to an articulate set whether or not they have ever been actually scored. But saying this is not enough either; for it is precisely in the context of discussions of conventionally scored programme music that one most frequently encounters talk of "musical painting", "representation", or "pictorial music". T o exclude these cases would be to eliminate the paradigms of musical representation, at least as reflected in the parlance of music theory and criticism. I n other words, an analysis of representational music should allow for it being so under standard notation. This appears to contradict Goodman's claims above, but I believe the contradiction can be eliminated by marshalling a distinction and eliciting an assumption. T h e distinction is one Goodman carefully draws earlier between a score and a notation. Notationality refers to the conjunction of five properties of a system: syntactic disjointness and articulateness, unambiguity, and semantic disjointness and articulateness. ([2], pp. 130-141 and 148-157) Any system lacking even one of these features thus far fails of notationality. Theoretically and practically, it is these five characteristics of a notational system that make possible the mutual recoverability of score and performance. I n fact, many modern scoring methods fall short of strict notationality (usually through violation of syntactic or semantic articulateness), and not even standard music notation is entirely "notational" inasmuch as it incorporates ambiguous symbols, some linguistic and some not. ([2], pp. 183-185) Various expression marks and verbal instructions regarding tempo, attack, and dynamics are ambiguous. I t is by virtue of just these nonnotational features of the music that a given passage may represent something. This is because the sound-compliants of these ambiguous elements of the score, when themselves taken as symbols, turn out to be both syntactically and semantically undifferentiated. Suppose, for instance, a few bars of stereotypical "galloping music" of the deedle-dum, deedle-dum, deedle-dum-dum-durn variety, duly written out in standard notation. Suppose further that the music indicates the approach or recession of a rider,2 A rider? Of what? A horse, camel, or ostrich ? Though ancillary clues as to what specific objects are denoted can be gleaned from context or accompanying titles, it may be thought that such questions constitute a peculiar threat to the
ON REPRESENTATIONAL MUSIC
47
each instant of sound indicating a relative location by an increase or decrease of volume. Depending upon the dynamics, the rider could be represented either as approaching and then receding or receding and approaching again, e.g., '< >' or '> <'. Thus, within the same work-class as defined by the score, there are performances which may represent quite different events. That is, we may have ambiguity within the same work-class-precisely what Goodman does not assume in the quoted passage above. Rather he assumes no ambiguity between score and compliant performance and no ambiguity between the music as sound event and what it denotes. So, given both that a passage denotes and that it belongs to an articulate scheme, it is descriptive. Wagner's system of leitmotifs contain convenient examples of musical description. Description with motifs is commonly a matter of class membership or predication as in natural languages. For instance, if a love motif be combined with a naming motif, the result is non-disjoint class inclusion or a musical description equivalent to saying so-and-so is in love. A motif may also be projected over a number of different characters or situations thereby achieving generality through having multiple compliants, and hence, descriptive status. T h e "Sturm-und-Drang Motiv" in Die Meistersinger appears to operate in this way as do group naming motifs like the "Meistersinger Motiv" which, among other things, can be construed as indicating class membership. I t might be objected that the scores of these examples contain ambiguous characters contrary to Goodman's assumption and should, therefore, qualify as representational rather than descriptive. This objection presupposes that music under a partially notational score is always representational if denotational. However, just as the mere presence or absence of a score does not mark the difference between the descriptive and the representational, neither does the mere presence of ambiguous parts of scores. Whether a denoting passage is representational depends upon whether the music denotes in respect of those aspects of the music which belong to an inarticulate scheme. That is not the case with the preceding examples from Wagner in which the non-notational aspects of the music contribute exclusively to its expressive -
-
denotative uses of music. However, this objection overlooks the fact that there is nothing intrinsic about the linguistic symbol "horse and rider" binding it inextricably to what in fact the phrase denotes. Nor are pictures (pictorial representations) any less subject to ambiguity and vagueness regarding what they denote, even, as we have seen, to the point of denoting things not pictured.
qualities. I n the example of the rider, however, the passage shows relative position-nearness or remoteness, approach or recessionby fluctuations in volume which are densely indicated in the score. Being therefore both syntactically and semantically dense with respect to the denoted event, the passage meets the requirements of a representational system. Besides this familiar device of silent film accompanists, there are less obvious examples of the use of music to construct a representational "sliding scale" in respect of some property or quality. As G. B. Shaw once observed, "It is a favorite trick of Wagner's, when one of his characters is killed on stage, to make the theme attached to that character weaken, fail, and fade away with a broken echo into silence." ([6], p. 106) Thus may a musical name be transformed to represent an event involving the character named.
5. System and Scheme Putting aside exceptions to notationality, a score in standard notation is a character in a notational system, the system consisting of scores and compliant performances. Where musical sounds are thought to have semantic (denotational) content, this "normal" situation is altered in at least two noteworthy respects: first, the sounds themselves acquire the syntactic features of a notational scheme, namely, syntactic disjointness and finite differentiation; but, secondly, scores and compliant performances become characters in different notational schemes. Allow me to explain this a bit further. Where a score S denoting sound events M constitutes a notational system, M as compliants of S must constitute a notational scheme when M is taken to denote something else X. This is because M retains the characteristics of disjointness and finite differentiation through the shift. as it were, from sound-compliant to sound-symbol. However, M must belong to a different notational scheme from S for the reason that X, the compliants of M , may not meet the semantic requirements of notationality. One is tempted to think of S and M under these circumstances on the analogy of written and verbal inscriptions in language where the printed or spoken word, say, 'cat' have a common compliant. But if it is assumed that S and M belong to the same notational scheme with respect to a common compliant X, then the notationality of the system S - M would be sacrificed in the event that X did not satisfy the semantic conditions of notationality. I n fact, the system M-X, as contrasted with the system S-M, may be either descriptive
ON REPRESENTATIONAL MUSIC
49
or notational depending upon whether the compliance classes of M are disjoint and articulate. I n language, both phonological elements and written inscriptions can be coextensive. I n music, scores clearly denote sounds, but taking scores or parts of them as denoting the compliants of the sounds would render scores nonnotational. I n other words, when used descriptively, musical sounds do not ascend (or descend, whichever one prefers) to the status of coordinate verbal utterances of written inscriptions as in language. However, as with language, a notational sound-scheme M is merely a necessary condition of description as well as of notationality, which leaves it open whether X will meet the semantic conditions of notationality or description. Nevertheless, there are practical circumstances where we might consolidate the characters of the score and sound schemes as belonging to one system. I n literary discussions of programme music and opera, for instance, one frequently encounters excerpted portions of a score labelled "The Siegfried Motif" or "The Sword Motif" being treated as if these score segments were coextensive with the musical sounds they denote. However, at most a score segment indirectly denotes a compliant of the music. I n the context of performance, it is a musical passage, not a score segment that denotes Siegfried. Temporarily, we may adopt the linguistic analogy for convenient reference; but only by ellipsis do a score S and compliant sounds A4 become coextensive inscriptions of the same characters.
6. Three Kinds of Mzlsical Denoting Musical naming and description are two kinds of musical denoting in which the sounds of music acquire syntactic characteristics relative to their denotata. T h e same applies also to musical representation as a third species of musical denoting. However, as I have argued, a distinguishing feature of representational music is that it denotes in respect of those musical aspects which are nonnotational, that is, do not belong to a notational scheme. T h e remainder of this paper will try to sharpen the differences between the three kinds of musical denoting and finally contrast them with musical expression as another kind of musical reference. Assume a notational scheme consisting of Figure 1 plus specific pitches and their durations-sounds-organized to satisfy the syntactic and semantic conditions of notationality. These sounds as characters, if taken to refer to something else so organized (say, different line lengths in inches) would then belong to a
notational scheme. T h e whole consisting of the notational scheme and the line intervals would constitute a notational system, in which particular pitches and lengths of line are mutually recoverable: F and 1 inch, G and 2 inches, and so on. At first glance, this manufactured example may seem remote from anything musically real. But consider that, among other things, it is the purpose of characters in a notational system such as a standard score to identify a work from one performance to another. ([2], p. 128) Similarly, a function of naming motifs in opera, so far as they are denotational, is to signal the identity of their compliants.
E denotes lines one inch long F denotes lines two inches long G denotes lines three inches long
Naming motifs are parts of a work denoting extra-musical things and however else they may function in the work, denotatively they serve mainly to identify and signalize their compliants whether particular objects, a character, group, or even as naming an event or feeling. As such, naming motifs constitute a redundant3 notational subsystem of a larger descriptive system of motifs. T h e only significant difference between Figure 1 and actual naming motifs from the standpoint of the symbol relations involved is that instead of individual pitches being correlated with different lengths of line, a whole musical passage is correlated with some item in a musicdrama. The passage may also be distinctive in its auditory appeal and expressive of qualities of its compliant, but these features should not be confused with its denotative use as a musical name. From the fact that musical sounds may be so organized as to satisfy the syntactic requirements of a notational scheme, it does not follow that they satisfy the semantic requirements of a notational system. As in Figure 2 (a) the sounds may denote overSince there
different variations of the same motif.
ON REPRESENTATIONAL MUSIC
51
lapping compliance classes thus sacrificing semantic disjointness and the mutual recoverability of sounds and compliants. Or they may denote ambiguously as in 2 (b). Either way, the sounds, while continuing to be parts of a notational scheme, belong to a descrip-
(a) E denotes lines between one and three inches long F denotes lines between two and four inches long G denotes lines between three and five inches long (b) E denotes lines one or two inches long
F denotes lines two or three inches long
G denotes lines three or four inches long
tive rather than notational system with respect to the extra-musical things they denote. If we introduce a slur (Figure 3) without finite differentiation between distinct pitches, and further stipulate that any difference in pitch denotes a difference in line length, the
system becomes both syntactically and semantically dense-a nonnotational scheme correlated with an undifferentiated field of reference-in short, a representational system not unlike the examples of the horseman and death motifs above.
7. Expression T h e topic of expression is introduced here more as an appendix than an extension of the preceding discussion. T h e three kinds of
musical denoting analysed above are thrown into some relief by comparing them to expression understood as a species of nondenoting musical reference. According to Goodman, what a work of art-in this instance a musical work-expresses are certain metaphorical labels denoting and exemplified by the music (121, pp. 85-95; see also 141). Naming, description, and representation depend upon what music under certain syntactic and semantic conditions is taken to denote. One difference, then, between expression as contrasted with naming, description, and representation is the difference between being denoted and denoting-denoting of versus denoting by music. Given that difference, expression is a matter of the metaphorical exemplification of properties belonging to the music solely as music. Since expression is not conditional upon the music's fulfilling certain syntactic requirements or its denoting anything, expression may characterize denoting and non-denoting music either as an additional nuance of musical names, descriptions, and representations, or as the sole extra-musical reference. Many a musical "programme" so-called is purely expressive. Beethoven, for instance, described his Pastoral Symphony as "expression of feeling rather than painting" ([I], p. 649)-the point being that it was in no way intended to denote any actual or generalized scene. Much the same thing could be said of the greater portion of Romantic music, namely that it is metaphorically exemplificational (expressive) rather than denotational, i.e., descriptive or representational. As I have argued elsewhere:
. . . (M)usic may metaphorically exemplify many of the labels literally applicable to, say, a love scene (e.g., tender, pleading, yielding, passionate) and thereby expressively underscore a dramatic action without necessarily denoting anything. Much programme music which seems prinzafacie descriptive, actually belongs in the . . . category of expressive music; which is to say that the music exemplifies literally or metaphorically many of the same labels as denote the dramatic action or whatever is mentioned in the title. Of course, descriptive music is usually also expressive, but the reverse generalization is less likely to hold. . . . (F)requently we are at a loss to decide whether a piece is denotational. A proper performance of Debussy's La iMer, for instance, may not so much describe the sea as express qualities of it, many of which belong literally to the sea and metaphorically to the music, e.g., shimmering, swirling, heaving, swelling, ebbing, flowing, gurgling. However, any number of other things might exhibit the same or similar properties. Had the title of Debussy's work been iWal de Mer, most of the same labels would continue to apply, in-
ON REPRESENTATIONAL MUSIC
53
cluding even the titles of the subsections (e.g., "De L'Aube a midi sur la mer") transformed thereby into hideous jokes! There are of course as many similarities between the sea, Debussy's music, and nausea as there are labels applicable to all three; but it requires no attention to what, if anything, the music denotes to discover any of them. On the other hand, there is nothing to prevent anyone from taking Debussy's piece as descriptive of the sea, and there are characteristics of the work which encourage that interpretation. The subtitles of each symphonic poem not only set an expressive mood but delimit specific aspects of the sea: "jeux de vagues", "Dialogue du vent et de la mer". Moreover, that the music expresses qualities of the sea in no way prohibits its also describing the sea provided certain syntactic and semantic conditions are met. T h e point is that the music can be so construed, and if La Mer denotes, it does so descriptively. [5].
I n conclusion, I have tried to show how music can be representational by comparison to the ways in which it may also be descriptive or expressive. Beyond a few hints limited to particular instances, no general criteria or tests for detecting what a piece represents, names, describes, or expresses have been suggested. Rather, it was argued that if music represents or describes, it must satisfy certain syntactic and semantic conditions which are the analogues of conditions more typical of pictorial and linguistic systems. Neither were any suggestions made for deciding the relative merits of the referential uses of music. T h e preceding analyses are compatible with all artistic attitudes from reverence to revulsion for programme devices. Instead, focus has been on certain symbol relations, however one prefers to name them, which are characteristic of some music and have their counterparts in other symbol systems.
[I] Apel, Willi. Harvard Dictionary of Music, 2nd ed., Cambridge, Mass., 1969. [2] Goodman, Nelson. Languages of A r t , Indianapolis, 1968. [31 . "Seven Strictures on Similarity," in Experience and Theory, ed. by Lawrence Foster and J. W. Swanson, Amherst, 1970, 19-29. 141 Howard, V. A. "On Musical Expression," T h e British Journal of Aesthetics, 11, 268-280. [51 . "On Musical Denoting," unpublished paper. 161 Shaw, G. B. T h e Perfect Wagnerite, New York, 1967.